Analyzing spatial characters of the ECG signal via complex network method

In recent years, various nonlinear time series analysis methodologies have been applied to study cardiac arrhythmias electrocardiograph (ECG) signals. They exhibit competitive advantages in comparison with the conventional linear methods but most of them heavily rely on the phase space reconstruction. In this paper the arrhythmias electrocardiograph (ECG) signals is investigated from the perspective of networks so as to find another effective approach independent of the phase space reconstruction to deeply discover some unknown information underlying the signals. The ECG signal is thereby transformed to a network topology, where its R-R cycles are regarded as nodes in the network, and link weights between two nodes are determined by Euclidean distance of corresponding two cycles. We then employ the network statistical criteria to discover the distinction among different cardiac rhythms. We validate this idea with atrial fibrillation (AF) and normal sinus rhythm (NSR) ECG signals. The results demonstrate that the differences between them can be well revealed from this novel perspective. The described method provides an insight into cardiac arrhythmias studies.